Analysis of rare variants requires statistical methods that are fundamentally different from association statistics used for testing common variants. There are two reasons for this. First, rare variants have to be combined in a gene (or pathway) for an association test to reach sufficient power57. For example, a causal SNP at a frequency of 1 in 500 and genotype relative risk of 10 in a sample of 200 cases and 200 controls, has 0.2% power to be detected at a conventional significance threshold for GWAS (P < 5 × 10−8). Second, functional and population genetics information can be added to the testing approach because exome sequencing comprehensively captures variation that can be annotated with such information.
